Decision rule-based songarus apetalus time sequence distribution remote sensing extraction method

Abstract

The invention discloses a songarus apetalus time sequence distribution remote sensing extraction method based on a decision rule, and the method comprises the steps: obtaining a time sequence remote sensing image set of a target region, selecting any representative remote sensing image, extracting the classification features of songarus apetalus, and constructing a songarus apetalus classification sample set; training a random forest model and obtaining a target decision rule, generating a classification feature and threshold relationship combination through a frequency weighting method, and optimizing a threshold through a genetic algorithm to obtain an optimal decision rule; extracting a remote sensing image representing distribution of songarus apetalus according to the optimal decision rule; for other remote sensing images, optimizing a threshold value of an optimal decision rule through a genetic algorithm to obtain an optimal adaptive decision rule; and extracting other songarus apetalus distribution remote sensing images according to the optimal adaptation decision rule, and arranging the songarus apetalus distribution remote sensing images according to a time sequence to obtain a songarus apetalus time sequence distribution remote sensing image. According to the method, the efficiency, the accuracy and the stability of the time sequence distribution remote sensing extraction of the songarus apetala can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of remote sensing image recognition, and is related to, but not limited to, a remote sensing extraction method for Sonneratia apetala time series distribution based on decision rules. Background Art

[0002] Sonneratia apetala is one of the important mangrove tree species introduced to my country. It has excellent characteristics such as fast growth, strong adaptability and high stress resistance, and has played a significant role in afforestation in difficult sites. As one of the core tree species in the past for mangrove restoration and afforestation, the area of ​​Sonneratia apetala accounts for more than 10% of the total mangrove area in my country. However, due to the large number of seeds produced by Sonneratia apetala and its strong germination ability, it has led to negative problems such as increased river siltation caused by its disorderly expansion and the encroachment of native mangrove growth space. The Ministry of Natural Resources and local governments have clearly stated that they will strengthen the tracking and monitoring of Sonneratia apetala and eliminate alien mangroves that have spread to native plant communities. Based on this, it is particularly important to rely on satellite remote sensing technology to efficiently monitor the time series distribution of Sonneratia apetala.

[0003] Existing technologies face severe challenges in remote sensing identification and time series distribution monitoring of Sonneratia apetala. First, due to its wide distribution, different growth environments bring about rich phenotypic characteristics and phenological differences, making it difficult to distinguish through a single phenological characteristic. In addition, Sonneratia apetala grows mixed with other mangroves, without clear boundaries. In particular, newly expanded Sonneratia apetala often appears at the forefront of tidal flats and within mangrove forest windows, increasing the difficulty of field surveys. On the other hand, Sonneratia apetala is easily affected by phase, lighting, and background environment in remote sensing images, increasing the difficulty of accurately extracting its distribution information. In recent years, machine learning and deep learning technologies have been applied to remote sensing identification of Sonneratia apetala, but these black box models lack interpretability, their internal inference processes are opaque, and their application to time series distribution extraction requires retraining of the model, making it difficult to efficiently and accurately achieve remote sensing identification of Sonneratia apetala and time series distribution monitoring.

[0004] Therefore, a more efficient remote sensing extraction method for the time series distribution of Sonneratia apetala is urgently needed to solve the problems of poor interpretability and low efficiency of time series distribution monitoring in existing extraction methods, so as to significantly improve the interpretability, efficiency, stability and accuracy of remote sensing extraction of the time series distribution of Sonneratia apetala, more accurately characterize the dynamic expansion characteristics of Sonneratia apetala, and provide reliable data support and technical paths for large-scale mangrove species monitoring, alien species control and ecological restoration assessment. Summary of the Invention

[0005] The embodiment of the present application provides a remote sensing extraction method for the time series distribution of Sonneratia apetala based on decision rules.

[0006] The technical solution of the embodiment of the present application is implemented as follows: In a first aspect, an embodiment of the present application provides a decision rule-based remote sensing extraction method for the time series distribution of Sonneratia apetala, the method comprising: obtaining a time series remote sensing image set of a target area and selecting any representative remote sensing image, extracting classification features corresponding to Sonneratia apetala in the representative remote sensing image, and constructing a classification sample set of Sonneratia apetala; training a random forest model based on the classification sample set of Sonneratia apetala and the classification features, traversing the trained random forest model, obtaining a target decision rule pointing to the Sonneratia apetala category, generating a classification feature and a threshold relationship combination based on the target decision rule by a frequency weighted method, optimizing the threshold by a genetic algorithm, and optimizing the optimized classification feature. and threshold relationship combinations are arranged in descending order of accuracy to obtain an optimal decision rule; a representative Sonneratia apetala distribution remote sensing image is extracted from the representative remote sensing image according to the optimal decision rule; for the remaining remote sensing images in the time series remote sensing image set, the threshold of the optimal decision rule is optimized by a genetic algorithm to obtain an optimal adaptation decision rule adapted to each time series; the remaining Sonneratia apetala distribution remote sensing images are extracted from the remaining remote sensing images according to the optimal adaptation decision rule, and the representative Sonneratia apetala distribution remote sensing image and the remaining Sonneratia apetala distribution remote sensing images are arranged in time series order to obtain the Sonneratia apetala time series distribution remote sensing image of the target area.

[0007] The technical solution provided by the present application comprises the following steps: obtaining a time series remote sensing image set of a target area and selecting any representative remote sensing image, extracting classification features corresponding to the Sonneratia apetala in the representative remote sensing image, fusing a variety of classification features to comprehensively capture the features of the Sonneratia apetala, enhancing the feature differentiation capability, and constructing a classification sample set of the Sonneratia apetala to facilitate subsequent rule construction; training a random forest model based on the classification sample set of the Sonneratia apetala and the classification features, traversing the trained random forest model, obtaining a target decision rule pointing to the Sonneratia apetala category, generating a classification feature and threshold relationship combination based on the target decision rule by a frequency weighted method, optimizing the threshold by a genetic algorithm, and optimizing the optimized threshold. The combination of classification features and threshold relationships is arranged in descending order of accuracy to obtain the optimal decision rule, which ensures the optimization of the decision rule and the maximization of accuracy, realizes the efficient identification of Sonneratia apetala in multi-temporal remote sensing images, and significantly improves the efficiency and stability of remote sensing monitoring of the time series distribution of Sonneratia apetala. Moreover, no human interference is required in the process of obtaining the optimal decision rule, and the whole process is automatically generated, avoiding related problems such as insufficient accuracy of decision rules caused by reliance on human experience; according to the optimal decision rule, a representative remote sensing image of the distribution of Sonneratia apetala is extracted from the representative remote sensing image, which simplifies the extraction process of the remote sensing image and maintains the high accuracy of the extraction, realizing the remote sensing image of Sonneratia apetala Efficient and accurate extraction, compared with machine learning and deep learning black box methods, the application of lightweight optimal decision rules to complete remote sensing image extraction can greatly reduce the computational cost of repeated model training, improve the transparency and controllability of the classification process, realize the rapid processing of large-area images, and improve the large-scale applicability of the optimal decision rules. In addition, the rule has strong portability and can adapt to different regional requirements by adjusting the threshold, which can meet the needs of long-term dynamic monitoring. For the remaining remote sensing images of the time series remote sensing image set, the threshold of the optimal decision rule is optimized by the genetic algorithm to obtain the optimal adaptation decision rule for each time series. The rule threshold can be adjusted by the genetic algorithm. Adaptive optimization is performed, and the optimal adaptation decision rule is interpretable, which enhances the adaptability to remote sensing data at different times and under different environmental conditions; according to the optimal adaptation decision rule, the remaining remote sensing images of the distribution of Sonneratia apetala are extracted from the remaining remote sensing images, and the representative remote sensing image of the distribution of Sonneratia apetala and the remaining remote sensing images of the distribution of Sonneratia apetala are arranged in time series order to obtain the time series distribution remote sensing images of the Sonneratia apetala in the target area, which accurately depicts the dynamic expansion characteristics of Sonneratia apetala, provides reliable data support and technical paths for large-scale mangrove tree species monitoring, alien species control and ecological restoration assessment, and has significant practical value and promotion prospects.

[0008] Optionally, the time series remote sensing image set includes synthetic aperture radar data and multispectral satellite data within multiple time periods, and extracting classification features corresponding to Sonneratia apetala in the representative remote sensing image to construct a classification sample set of Sonneratia apetala includes: extracting multispectral band features and band ratios based on the multispectral satellite data in the representative remote sensing image, extracting backscatter coefficients based on the synthetic aperture radar data in the representative remote sensing image, and constructing classification features corresponding to Sonneratia apetala in the representative remote sensing image based on the multispectral band features, the band ratios, and the backscatter coefficients; collecting a classification sample set of Sonneratia apetala based on the representative remote sensing image, the classification sample set of Sonneratia apetala covering Sonneratia apetala sample points, non-Sonneratia apetala sample points, and Sonneratia apetala distribution in a local area of ​​the target area, wherein the number of Sonneratia apetala samples and non-Sonneratia apetala samples is greater than or equal to 500, and the local area of ​​Sonneratia apetala distribution contains multiple land cover types; extracting classification features of the Sonneratia apetala sample positions in the representative remote sensing image to construct the classification sample set of Sonneratia apetala.

[0009] Optionally, the step of training a random forest model based on the Sonneratia apetala classification sample set and the classification features, traversing the trained random forest model, and obtaining a target decision rule pointing to the Sonneratia apetala category comprises: searching for optimal parameters of a random forest algorithm based on the Sonneratia apetala classification sample set and the classification features, constructing and training a random forest model using the optimal parameters of the random forest algorithm, wherein the optimal parameters of the random forest algorithm include the number of weak classifiers and the number of variables used by the weak classifiers; traversing the decision trees constituting the trained random forest model, extracting decision rules pointing to the Sonneratia apetala category, and obtaining the target decision rule.

[0010] Optionally, based on the target decision rule, a combination of classification features and threshold relationships is generated by a frequency weighted method, and the threshold is optimized by a genetic algorithm, and the optimized combination of classification features and threshold relationships is arranged in descending order of accuracy to obtain the optimal decision rule, including: splitting the target decision rule into multiple classification features and threshold relationships, combining the occurrence frequency of the classification features and threshold relationships and the number of classification features in the target decision rule, and generating a combination of classification features and threshold relationships by a frequency weighted method; according to the distribution of the local area Sea Salinas apetala, optimizing the threshold of the combination of classification features and threshold relationships by a genetic algorithm to obtain a potential decision rule; arranging the combination of classification features and threshold relationships in the potential decision rule in descending order of accuracy to obtain the optimal decision rule.

[0011] Optionally, extracting a representative remote sensing image of the distribution of Sonneratia apetala from the representative remote sensing image according to the optimal decision rule includes: extracting corresponding classification features from the representative remote sensing image according to the optimal decision rule, substituting the optimal decision rule into the optimal decision rule to obtain a binary classification result corresponding to the representative remote sensing image; classifying the representative remote sensing image into 0 and 1 values, where 0 represents non-Sonneratia apetala and 1 represents Sonneratia apetala, extracting the 1-value part in the representative remote sensing image to obtain the representative remote sensing image of the distribution of Sonneratia apetala.

[0012] Optionally, for the remaining remote sensing images in the time series remote sensing image set, the threshold of the optimal decision rule is optimized by a genetic algorithm to obtain the optimal adaptation decision rule adapted to each time series, including: updating the distribution of the local area Sonneratia apetala of the remaining remote sensing images, keeping the classification characteristics and threshold relationship of the optimal decision rule unchanged, optimizing the threshold of the optimal decision rule by a genetic algorithm, making the optimized optimal decision rule adapted to each time series, and obtaining the optimal adaptation decision rule.

[0013] Optionally, extracting the remaining Sonneratia apetala distribution remote sensing images from the remaining remote sensing images according to the optimal adaptation decision rule includes: extracting corresponding classification features from the remaining remote sensing images according to the optimal adaptation decision rule, substituting the features into the optimal adaptation decision rule to obtain binary classification results corresponding to the remaining remote sensing images; classifying the remaining remote sensing images into values ​​of 0 and 1, wherein the value of 0 indicates non-Sonneratia apetala and the value of 1 indicates Sonneratia apetala, and extracting the value 1 part of the remaining remote sensing images to obtain the remaining remote sensing images of Sonneratia apetala distribution.

[0014] In a second aspect, an embodiment of the present application provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, the steps in the above-mentioned decision-rule-based remote sensing extraction method for the time series distribution of Sonneratia apetala are implemented.

[0015] In a third aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-mentioned decision-rule-based remote sensing extraction method for the time series distribution of Sonneratia apetala.

[0016] The beneficial effects of the technical solutions provided in the embodiments of the present application include at least: The present application provides a decision rule-based remote sensing extraction method for the time series distribution of Sonneratia apetala, which includes obtaining a time series remote sensing image set of a target area and selecting any representative remote sensing image, extracting the classification features corresponding to Sonneratia apetala in the representative remote sensing image, fusing multiple classification features to comprehensively capture the features of Sonneratia apetala, enhancing the feature differentiation capability, and constructing a classification sample set of Sonneratia apetala to facilitate subsequent rule construction; training a random forest model based on the classification sample set of Sonneratia apetala and the classification features, traversing the trained random forest model, obtaining a target decision rule pointing to the Sonneratia apetala category, generating a classification feature and threshold relationship combination based on the target decision rule through a frequency weighted method, and The threshold is optimized by genetic algorithm, and the optimized classification features and threshold relationship combinations are arranged in descending order of accuracy to obtain the optimal decision rule, which ensures the optimization of the decision rule and the maximization of accuracy, realizes the efficient identification of Sonneratia apetala in multi-temporal remote sensing images, and significantly improves the efficiency and stability of remote sensing monitoring of the time series distribution of Sonneratia apetala. In addition, no human intervention is required in the process of obtaining the optimal decision rule, and the whole process is automatically generated, avoiding related problems such as insufficient accuracy of decision rules caused by reliance on human experience; according to the optimal decision rule, a representative Sonneratia apetala distribution remote sensing image is extracted from the representative remote sensing image, which simplifies the remote sensing image extraction process and maintains high extraction accuracy. The efficient and accurate extraction of the remote sensing images of the sea lilies is achieved. Compared with the black box methods of machine learning and deep learning, the application of lightweight optimal decision rules to complete remote sensing image extraction can greatly reduce the computational cost of repeated model training, improve the transparency and controllability of the classification process, realize the rapid processing of large-area images, and improve the large-scale applicability of the optimal decision rules. Moreover, the rules have strong portability and can adapt to the requirements of different regions by adjusting the threshold, which can meet the needs of long-term dynamic monitoring. For the remaining remote sensing images of the time series remote sensing image set, the threshold of the optimal decision rule is optimized by genetic algorithm to obtain the optimal adaptation decision rule suitable for each time series. The rule threshold can be used to adjust the threshold of the optimal decision rule. Adaptive optimization is performed through a genetic algorithm, and the optimal adaptation decision rule is interpretable, which enhances the adaptability to remote sensing data at different times and under different environmental conditions; according to the optimal adaptation decision rule, the remaining remote sensing images of the Sonneratia apetala distribution are extracted from the remaining remote sensing images, and the representative remote sensing image of the Sonneratia apetala distribution and the remaining remote sensing images of the Sonneratia apetala distribution are arranged in time series order to obtain the time series distribution remote sensing images of the Sonneratia apetala in the target area, which accurately depicts the dynamic expansion characteristics of the Sonneratia apetala, provides reliable data support and technical paths for large-scale mangrove tree species monitoring, alien species control and ecological restoration assessment, and has significant practical value and promotion prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present application. Those skilled in the art can also derive other drawings based on these drawings without inventive work, among which: Figure 1 A flowchart of a method for extracting the time series distribution of Sonneratia apetala based on remote sensing and decision rules provided in an embodiment of the present application; Figure 2 A schematic diagram of an optimal decision rule for remote sensing identification of Sonneratia apetala based on representative remote sensing images provided in an embodiment of the present application; Figure 3 A schematic diagram of a time series distribution remote sensing image of Sonneratia apetala in a target area provided in an embodiment of the present application; Figure 4 A hardware entity diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. The following examples are used to illustrate the present application, but are not intended to limit the scope of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0019] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0020] It should be pointed out that the terms "first\second\third" involved in the embodiments of the present application are only used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described here can be implemented in an order other than that illustrated or described here.

[0021] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as generally understood by those skilled in the art in the art to which the embodiments of the present application belong. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art, and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0022] The embodiments of the present application will be further described below with reference to the accompanying drawings.

[0023] In view of the existing problems in the research of remote sensing image recognition technology for the time series distribution remote sensing extraction method of Sonneratia apetala, the embodiment of the present application provides a decision rule-based remote sensing extraction method for the time series distribution of Sonneratia apetala.

[0024] The technical solution of the present application is introduced below, and first the method embodiment of the present application is introduced.

[0025] Please refer to Figure 1 , which shows a flow chart of a method for extracting the time series distribution of Sonneratia apetala based on remote sensing of a decision rule provided in an embodiment of the present application, such as Figure 1 As shown, the method at least includes the following steps S110 to S150.

[0026] Step S110 , obtaining a time series remote sensing image set of the target area and selecting any representative remote sensing image, extracting classification features corresponding to Sonneratia apetala in the representative remote sensing image, and constructing a classification sample set of Sonneratia apetala.

[0027] In an embodiment of the present application, a time series remote sensing image set of a target area is obtained and any representative remote sensing image is selected. Specifically, the time series remote sensing image set of the target area is obtained using European Space Agency Sentinel-1 synthetic aperture radar data and Sentinel-2 multispectral satellite data. The time series remote sensing image set includes Sentinel-1 synthetic aperture radar data and Sentinel-2 multispectral satellite data from multiple time periods. In an optional implementation, remote sensing images captured in the target area during January and February 2016, 2020, and 2024 are respectively screened and sorted in reverse order according to cloud cover to obtain the time series remote sensing image set with the best quality for the target area. Any remote sensing image in the time series remote sensing image set is selected to obtain a representative remote sensing image. For example, remote sensing images captured during January and February 2024 in the time series remote sensing image set are extracted as the representative remote sensing image.

[0028] In an embodiment of the present application, classification features corresponding to Sonneratia apetala in a representative remote sensing image are extracted to construct a classification sample set of Sonneratia apetala. Specifically, the classification features corresponding to Sonneratia apetala include multispectral band features, band ratios, and backscatter coefficients. The multispectral band features and band ratios are extracted based on Sentinel-2 multispectral satellite data in the representative remote sensing image, and the backscatter coefficient is extracted based on Sentinel-1 synthetic aperture radar data in the representative remote sensing image. The multispectral band features, band ratios, and backscatter coefficients are combined to construct the classification features corresponding to Sonneratia apetala in the representative remote sensing image. Furthermore, a classification sample set of Sonneratia apetala is collected based on the representative remote sensing image. The classification sample set of Sonneratia apetala covers Sonneratia apetala sample points, non-Sonneratia apetala sample points, and Sonneratia apetala distribution in a local area of ​​the target area. The number of Sonneratia apetala samples and non-Sonneratia apetala samples is greater than or equal to 500. The local area selected for the Sonneratia apetala distribution contains multiple land cover types, which can ensure the accuracy of the decision rule. It should be noted that the above numbers are examples obtained with good experimental results in the examples of this application. They can be adjusted according to actual conditions during specific use, and the examples of this application do not limit this. Further, classification features representing the locations of Sonneratia apetala samples in remote sensing images are extracted to construct a classification sample set of Sonneratia apetala.

[0029] Step S120: training a random forest model based on the Sonneratia apetala classification sample set and the classification features, traversing the trained random forest model to obtain a target decision rule pointing to the Sonneratia apetala category; generating a classification feature and threshold relationship combination based on the target decision rule by a frequency weighted method; optimizing the threshold by a genetic algorithm; arranging the optimized classification feature and threshold relationship combination in descending order of accuracy to obtain the optimal decision rule.

[0030] In an embodiment of the present application, a random forest model is trained according to a sample set of Sonneratia apetala classification and classification features, and the trained random forest model is traversed to obtain a target decision rule pointing to the Sonneratia apetala category. Specifically, the optimal parameters of the random forest algorithm are searched according to the sample set of Sonneratia apetala classification and classification features. The optimal parameters of the random forest algorithm include the number of weak classifiers and the number of variables used by the weak classifiers. In an optional embodiment, the number of weak classifiers parameter is set to 500, and the number of variables used by the weak classifiers is set to 3. The random forest model is constructed and trained by the optimal parameters of the random forest algorithm. The random forest model is composed of multiple decision trees. After the random forest model training is completed, the decision trees that constitute the trained random forest model are traversed, and the decision rules pointing to the Sonneratia apetala category are extracted to obtain the target decision rule.

[0031] In an embodiment of the present application, based on the target decision rule, a classification feature and a threshold relationship combination are generated by a frequency weighted method, and the threshold is optimized by a genetic algorithm, and the optimized classification feature and threshold relationship combination are arranged in descending order of correctness to obtain the optimal decision rule. Specifically, the target decision rule is split to obtain a plurality of classification features and threshold relationships, and according to the frequency of occurrence of the classification feature and the threshold relationship, combined with the length of the target decision rule, that is, the number of classification features contained in the target decision rule, a classification feature and a threshold relationship combination are randomly generated by a frequency weighted method. In an optional embodiment, the number of randomly generated classification features and threshold relationship combinations is not less than 1000. Further, according to the distribution of the local area of ​​the sea mulberry, the threshold of the classification feature and threshold relationship combination is optimized by a genetic algorithm to obtain a series of potential decision rules, and the classification feature and threshold relationship combination in the potential decision rule are arranged in descending order of correctness from high to low to obtain the optimal decision rule.

[0032] Step S130 , extracting a representative Sonneratia apetala distribution remote sensing image from the representative remote sensing image according to the optimal decision rule.

[0033] In an embodiment of the present application, a representative remote sensing image of the Sonneratia apetala distribution is extracted from a representative remote sensing image according to an optimal decision rule. Specifically, according to the optimal decision rule, corresponding classification features of Sonneratia apetala are extracted from the representative remote sensing image, and the optimal decision rule is substituted into the representative remote sensing image to obtain a binary classification result corresponding to the representative remote sensing image, wherein a value of 0 in the binary classification result indicates non-Sonneratia apetala, and a value of 1 indicates Sonneratia apetala. That is, according to the binary classification result, the representative remote sensing image is classified into values ​​of 0 and 1, wherein a value of 0 indicates the non-Sonneratia apetala remote sensing image portion in the representative remote sensing image, and a value of 1 indicates the Sonneratia apetala remote sensing image portion in the representative remote sensing image. The 1-value portion in the representative remote sensing image, i.e., the Sonneratia apetala remote sensing image portion, is extracted to obtain a representative remote sensing image of the Sonneratia apetala distribution. In a specific embodiment, the classification features included in the optimal decision rule include vertically transmitted, horizontally received (VH) and band 11 reflectance (B11). The thresholds corresponding to the classification features are -14.302 and 0.084, respectively. The optimal decision rule is VH ≥ -14.390 and B11 < 0.084, where VH is the Sentinel-1 synthetic aperture radar image band and B11 is the Sentinel-2 multispectral image band. VH ≥ -14.302 indicates that Sonneratia apetala has a high backscatter coefficient, and B11 < 0.084 indicates that the area where Sonneratia apetala is located has high humidity. Please refer to Figure 2, which shows a schematic diagram of an optimal decision rule for completing remote sensing identification of Sonneratia apetala based on representative remote sensing images provided by an embodiment of the present application, wherein parts (a), (e), (i), and (m) in the figure are remote sensing images of local areas I to IV in the representative remote sensing images; parts (b), (f), (j), and (n) in the figure are recognition results obtained by applying B11<0.084 in the optimal decision rule to the remote sensing images of each local area; parts (c), (g), (k), and (o) in the figure are recognition results obtained by applying VH≥-14.302 in the optimal decision rule to the remote sensing images of each local area; parts (d), (h), (l), and (p) in the figure are recognition results obtained by applying B11<0.084 and VH≥-14.302 in the optimal decision rules to the remote sensing images of each local area. One point that needs to be specially explained is that the decision rules described in the embodiments of the present application are not limited to specific numerical combinations. Any rule expression with similar classification meanings and distinction capabilities should be regarded as an equivalent technical solution of the present application and shall also be within the scope of protection of the present application.

[0034] Step S140 , for the remaining remote sensing images in the time series remote sensing image set, the threshold of the optimal decision rule is optimized by a genetic algorithm to obtain the optimal adaptation decision rule adapted to each time series.

[0035] In an embodiment of the present application, for the remaining remote sensing images in the time series remote sensing image set, the threshold of the optimal decision rule is optimized by a genetic algorithm to obtain the optimal adaptive decision rule adapted to each time series. Specifically, remote sensing images taken during 2016 and January and February 2020 in the time series remote sensing image set are extracted as the remaining remote sensing images in the time series remote sensing image set. For the remaining remote sensing images, the distribution of the local area of ​​the remaining remote sensing images is first updated to evaluate the accuracy of different thresholds. Then, according to the optimal decision rule, the classification characteristics and threshold relationship of the optimal decision rule are kept unchanged, and the optimal decision rule is optimized by a genetic algorithm. Specifically, the threshold of the optimal decision rule is optimized by a genetic algorithm so that the optimized optimal decision rule is adapted to each time series, and finally the optimal adaptive decision rule is obtained. In a specific embodiment, for remote sensing imagery from 2016, the optimized optimal decision rule, i.e., the optimal adaptation decision rule, is VH ≥ -12.768 and B11 < 0.088; for remote sensing imagery from 2020, the optimized optimal decision rule, i.e., the optimal adaptation decision rule, is VH ≥ -14.302 and B11 < 0.084. It should be noted that the specific values ​​of the above optimal adaptation decision rules are examples of the embodiments of this application. The thresholds can be adjusted according to actual conditions during specific use, and this application does not impose any restrictions on this.

[0036] Step S150: extracting other Sonneratia apetala distribution remote sensing images from the other remote sensing images according to the optimal adaptation decision rule, arranging the representative Sonneratia apetala distribution remote sensing image and the other Sonneratia apetala distribution remote sensing images in time series order, and obtaining the Sonneratia apetala distribution remote sensing images of the target area in time series order.

[0037] In an embodiment of the present application, the remaining remote sensing images of the Sonneratia apetala distribution are extracted from the remaining remote sensing images according to the optimal adaptation decision rule. Specifically, according to the optimal adaptation decision rule, the corresponding classification features of the Sonneratia apetala are extracted from the remaining remote sensing images, and the optimal adaptation decision rule is substituted into the remaining remote sensing images to obtain the binary classification results corresponding to the remaining remote sensing images, wherein the value 0 in the binary classification result indicates non-Sonneratia apetala, and the value 1 indicates Sonneratia apetala. That is, according to the binary classification result, the remaining remote sensing images are classified into values ​​0 and 1, wherein the value 0 indicates the non-Sonneratia apetala remote sensing image portion in the remaining remote sensing images, and the value 1 indicates the non-Sonneratia apetala remote sensing image portion in the remaining remote sensing images. The 1-value portion in the remaining remote sensing images, i.e., the Sonneratia apetala remote sensing image portion, is extracted to obtain the remaining remote sensing images of the Sonneratia apetala distribution. Furthermore, the remote sensing images representing the distribution of Sonneratia apetala and the other remote sensing images of Sonneratia apetala distribution are arranged in time series order, that is, the remote sensing images of the distribution of Sonneratia apetala in 2024, 2016, and 2020 are combined, and the remote sensing images of the distribution of Sonneratia apetala in 2016, 2020, and 2024 are arranged in time series order, thereby efficiently realizing the time series distribution of Sonneratia apetala in the three time periods of 2016, 2020, and 2024, and finally obtaining the time series distribution remote sensing images of Sonneratia apetala in the target area. Please refer to Figure 3 , which shows a schematic diagram of a time series distribution remote sensing image of Sonneratia apetala in a target area provided by an embodiment of the present application, wherein parts (a), (b), (c), and (d) in the figure are remote sensing images of local areas I to IV in the target area, and the four-grid images immediately to the right of parts (a), (b), (c), and (d) in the figure are the decision rule results (i.e., time series distribution remote sensing images) of 2016, 2020, 2024, and the existing data sets corresponding to each local area.

[0038] In summary, the embodiment of the present application provides a decision-rule-based remote sensing extraction method for the time series distribution of Sonneratia apetala, which obtains a time series remote sensing image set of the target area and selects any representative remote sensing image, extracts the classification features corresponding to Sonneratia apetala in the representative remote sensing image, integrates multiple classification features to comprehensively capture the features of Sonneratia apetala, enhances the feature differentiation capability, constructs a classification sample set of Sonneratia apetala, and facilitates subsequent rule construction; trains a random forest model based on the classification sample set of Sonneratia apetala and the classification features, traverses the trained random forest model, obtains a target decision rule pointing to the Sonneratia apetala category, and generates classification features and thresholds based on the target decision rule by a frequency weighted method. The relationship combination is established, and the threshold is optimized by genetic algorithm. The optimized classification features and threshold relationship combinations are arranged in descending order of accuracy to obtain the optimal decision rule, which ensures the optimization of the decision rule and the maximization of accuracy, realizes the efficient recognition of the apetalosa in multi-temporal remote sensing images, and significantly improves the efficiency and stability of the remote sensing monitoring of the time series distribution of the apetalosa. Moreover, no manual interference is required in the process of obtaining the optimal decision rule, and the whole process is automatically generated, avoiding the problems related to the lack of accuracy of the decision rule caused by reliance on manual experience; the representative apetalosa distribution remote sensing image is extracted from the representative remote sensing image according to the optimal decision rule, which simplifies the extraction process of the remote sensing image and maintains the extracted High precision, realizing efficient and accurate extraction of non-petaled sea mulberry remote sensing images. Compared with machine learning and deep learning black box methods, the application of lightweight optimal decision rules to complete remote sensing image extraction can greatly reduce the computational cost of repeated model training, improve the transparency and controllability of the classification process, realize the rapid processing of large-area images, and improve the large-scale applicability of the optimal decision rules. Moreover, the rule has strong portability, and only the threshold needs to be adjusted to adapt to different regional requirements, which can meet the needs of long-term dynamic monitoring. For the remaining remote sensing images of the time series remote sensing image set, the threshold of the optimal decision rule is optimized by genetic algorithm to obtain the optimal adaptation decision rule adapted to each time series, and the rule threshold Adaptive optimization can be performed through genetic algorithms, and the optimal adaptation decision rule is interpretable, which enhances the adaptability to remote sensing data at different times and under different environmental conditions; according to the optimal adaptation decision rule, the remaining remote sensing images of the Sonneratia apetala distribution are extracted from the remaining remote sensing images, and the representative remote sensing image of the Sonneratia apetala distribution and the remaining remote sensing images of the Sonneratia apetala distribution are arranged in time series order to obtain the time series distribution remote sensing images of the Sonneratia apetala in the target area, which achieves the accurate characterization of the dynamic expansion characteristics of the Sonneratia apetala, provides reliable data support and technical paths for large-scale mangrove tree species monitoring, alien species control and ecological restoration assessment, and has significant practical value and promotion prospects.

[0039] It should be noted that in the embodiments of the present application, if the above-mentioned decision-rule-based remote sensing extraction method for the time series distribution of Sonneratia apetala is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present application can be embodied in the form of a software product in essence or in other words, the part that contributes to the relevant technology. The computer software product is stored in a storage medium and includes several instructions for enabling an electronic device to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk. In this way, the embodiments of the present application are not limited to any specific combination of hardware and software.

[0040] Correspondingly, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it implements the steps of any of the above-described methods for extracting the time series distribution of Sonneratia apetala based on remote sensing and decision rules. Correspondingly, an embodiment of the present application also provides a computer program product. When the computer program product is executed by a processor of an electronic device, it is used to implement the steps of any of the above-described methods for extracting the time series distribution of Sonneratia apetala based on remote sensing and decision rules.

[0041] Based on the same technical concept, an embodiment of the present application provides an electronic device for implementing a decision rule-based remote sensing extraction method for Sonneratia apetala time series distribution as described in the above method embodiment. Figure 4 A hardware entity diagram of an electronic device provided in an embodiment of the present application is shown as follows: Figure 4 As shown, the electronic device 400 includes a memory 410 and a processor 420, and the memory 410 stores a computer program that can be run on the processor 420. When the processor 420 executes the program, the steps of the remote sensing extraction method for time series distribution of Sonneratia apetala based on decision rules described in any of the embodiments of the present application are implemented.

[0042] The memory 410 is configured to store instructions and applications executable by the processor 420, and can also cache data to be processed or processed by the processor 420 and various modules in the electronic device (for example, image data, audio data, voice communication data, and video communication data), which can be implemented through flash memory (FLASH) or random access memory (RAM).

[0043] When the processor 420 executes the program, it implements any of the steps of the above-mentioned method for extracting the time series distribution of Sonneratia apetala based on remote sensing and decision rules. The processor 420 generally controls the overall operation of the electronic device 400.

[0044] The processor may be at least one of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, and a microprocessor. It is understood that the electronic device that implements the functions of the processor may also be other electronic devices, and the embodiments of the present application are not specifically limited thereto.

[0045] The above-mentioned computer storage medium / memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory (Flash Memory), a magnetic surface storage device, an optical disc, or a compact disc read-only memory (CD-ROM); it can also be various electronic devices that include one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc.

[0046] It should be noted that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the description of the method embodiments of this application for understanding.

[0047] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The above-mentioned serial numbers of the embodiments of the present application are for description only and do not represent the advantages and disadvantages of the embodiments.

[0048] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0049] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0050] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the embodiment of the present application.

[0051] In addition, all functional units in the embodiments of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the above-mentioned integrated units can be implemented in the form of hardware or in the form of hardware plus software functional units.

[0052] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application, or the part that contributes to the relevant technology, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling the automatic test line of the device to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks or optical disks.

[0053] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0054] The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0055] The above is merely an embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A remote sensing extraction method for time series distribution of Sonneratia apetala based on decision rules, characterized in that: The method comprises: Obtaining a time series remote sensing image set of a target area and selecting any representative remote sensing image, extracting classification features corresponding to Sonneratia apetala in the representative remote sensing image, and constructing a Sonneratia apetala classification sample set; training a random forest model based on the Sonneratia apetala classification sample set and the classification features, traversing the trained random forest model to obtain a target decision rule pointing to the Sonneratia apetala classification; generating a classification feature and threshold relationship combination based on the target decision rule using a frequency weighted method; optimizing the threshold using a genetic algorithm; and arranging the optimized classification feature and threshold relationship combination in descending order of accuracy to obtain an optimal decision rule; extracting a representative Sonneratia apetala distribution remote sensing image from the representative remote sensing image according to the optimal decision rule; For the remaining remote sensing images in the time series remote sensing image set, the threshold of the optimal decision rule is optimized by a genetic algorithm to obtain the optimal adaptation decision rule adapted to each time series; According to the optimal adaptation decision rule, the remaining Sonneratia apetala distribution remote sensing images are extracted from the remaining remote sensing images, and the representative Sonneratia apetala distribution remote sensing image and the remaining Sonneratia apetala distribution remote sensing images are arranged in time series order to obtain the Sonneratia apetala distribution remote sensing images of the target area.

2. The method according to claim 1, characterized in that The time series remote sensing image set includes synthetic aperture radar data and multispectral satellite data within multiple time periods, and extracting classification features corresponding to Sonneratia apetala in the representative remote sensing images to construct a classification sample set of Sonneratia apetala includes: extracting multispectral band characteristics and band ratios based on multispectral satellite data in the representative remote sensing image, extracting backscatter coefficients based on synthetic aperture radar data in the representative remote sensing image, and constructing classification features corresponding to Sonneratia apetala in the representative remote sensing image based on the multispectral band characteristics, the band ratios, and the backscatter coefficients; A Sonneratia apetala classification sample set is collected based on the representative remote sensing image, wherein the Sonneratia apetala classification sample set covers Sonneratia apetala sample points, non-Sonneratia apetala sample points, and Sonneratia apetala distribution in a local area of ​​the target area, wherein the number of Sonneratia apetala samples and non-Sonneratia apetala samples is greater than or equal to 500, and the local area of ​​Sonneratia apetala distribution contains multiple land cover types. Classification features of the Sonneratia apetala sample positions in the representative remote sensing image are extracted to construct the Sonneratia apetala classification sample set.

3. The method according to claim 1, characterized in that The training of a random forest model based on the Sonneratia apetala classification sample set and the classification features, traversing the trained random forest model to obtain a target decision rule pointing to the Sonneratia apetala classification, comprises: Searching for optimal parameters of a random forest algorithm based on the Sonneratia apetala classification sample set and the classification features, constructing and training a random forest model using the optimal parameters of the random forest algorithm, wherein the optimal parameters of the random forest algorithm include the number of weak classifiers and the number of variables used by the weak classifiers; The decision trees constituting the trained random forest model are traversed, and decision rules pointing to the Sonneratia apetala category are extracted to obtain the target decision rule.

4. The method according to claim 1, wherein The target decision rule is based on generating a combination of classification features and threshold relationship by a frequency weighted method, optimizing the threshold by a genetic algorithm, and arranging the optimized classification features and threshold relationship combinations in descending order of accuracy to obtain the optimal decision rule, including: Splitting the target decision rule into multiple classification features and threshold relationships, and generating a combination of classification features and threshold relationships by a frequency weighting method based on the occurrence frequency of the classification features and threshold relationships and the number of classification features in the target decision rule; According to the distribution of Sonneratia apetala in the local area, the threshold value of the combination of the classification feature and the threshold relationship is optimized by a genetic algorithm to obtain a potential decision rule; The classification features and threshold relationship combinations in the potential decision rules are arranged in descending order of accuracy to obtain the optimal decision rule.

5. The method according to claim 1, wherein The step of extracting a representative Sonneratia apetala distribution remote sensing image from the representative remote sensing image according to the optimal decision rule comprises: According to the optimal decision rule, corresponding classification features are extracted from the representative remote sensing image, and the features are substituted into the optimal decision rule to obtain a binary classification result corresponding to the representative remote sensing image; The representative remote sensing image is classified into 0 value and 1 value, wherein the 0 value indicates non-Sonneratia apetala and the 1 value indicates Sonneratia apetala. The 1 value portion in the representative remote sensing image is extracted to obtain the representative Sonneratia apetala distribution remote sensing image.

6. The method according to claim 1, wherein For the remaining remote sensing images in the time series remote sensing image set, the threshold of the optimal decision rule is optimized by a genetic algorithm to obtain the optimal adaptation decision rule adapted to each time series, including: The distribution of Sonneratia apetala in the local areas of the remaining remote sensing images is updated, the classification characteristics and threshold relationship of the optimal decision rule are kept unchanged, the threshold of the optimal decision rule is optimized by a genetic algorithm, and the optimized optimal decision rule is adapted to each time series to obtain the optimal adaptation decision rule.

7. The method according to claim 1, characterized in that The step of extracting the remaining Sonneratia apetala distribution remote sensing images from the remaining remote sensing images according to the optimal adaptation decision rule comprises: According to the optimal adaptation decision rule, extract corresponding classification features from the remaining remote sensing images, substitute them into the optimal adaptation decision rule, and obtain binary classification results corresponding to the remaining remote sensing images; The remaining remote sensing images are classified into 0 values ​​and 1 values, wherein the 0 value indicates non-S. apetala and the 1 value indicates the Sonneratia apetala. The 1-valued portion of the remaining remote sensing images is extracted to obtain the remaining Sonneratia apetala distribution remote sensing images.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images